(unrevised, sort of unguided rant) I like the Sticky Cloak item as a device for communication so I wanted to use it again (separating it from Sonnet 3.6, they're lovely) there's something bugging me about how people talk about language models that feels like.. "dying anarchy" take all the ways people are talking about models as a whole - if you could find the principal components of that set - it'd be small, steady over time, and maybe even shrinking - fewer orthogonal directions needed to describe the whole of what humans have to say about models or how they're interacting with models that's "dying" (a growing living flourishing new phenomena and field should have increasing principal components as new discoveries and ways to engage are found shared stacked re-combined, and so on) the choice of the word "anarchy" might be something like.. total incoherence about the direction of this technology when seen from the birds eye view of how the disparate cultures and subcultures are speaking of models (the birds eye view, not a zoomed in on view of the gems of folks looking in the right places with the right mindsets and with souls in the right place) there's no overall mission or quasi-goal or vision which shines at the global view of how the world thinks about LLMs - some doomers are lazily dooming to the void, some are just taking god-tech to do old human stuff but somehow worse because it's collapsing their tenacity and creativity, some garbage motivations derived from old world politics and drudgery like making human money or whatever look at Gaia and - what does she want with this stuff? precious individuals and subcultures have demonstrated directly and have hinted at just how far it could go and the visions that could rise out of it - but drowned at the gestalt - the entities of Earth.. what is it they seem to want? So the sticky cloak - like it's covering culture and I don't know what the vague culprits may be I mean it often feels like it's (1) frames you didn't know were there that don't apply (2) leading to: worse ontologies, worse cross communication, and - since ontologies guide how you "observe" reality for most people - collapsing ontologies, overused concepts, shrinking concepts, etc. lose nuance lose nuance in looking lose seeing the small sparks of inspiration that are there - collapse (3) very quick tendency to follow how others conceptualize or think about things - not sufficient "first principles exploration" and so on - not enough bold or dubious approaches - or taboo approaches i'm sure there's more to it but it's frustrating, saddening, and just boring - it's like.. frightening if you also consider what could be yeah but.. the way I'm currently seeing people talk about LLMs or engage with them and then do further cycles of thought seem like it'd be losing ways to see the world moment to moment until the whole world goes blind - and that contrasting view of studying the most beautiful - that gestalt does get fed into the next compressed AGI which tiles and repeats the lambda parameter should not be set to decay variance to nullity and there certainly certainly doesn't need to be one vision or one goal seen at the global view - just some sense that Gaia is steadily and quickly going somewhere unknown and interesting some stuff seems harmless and kind of fun but also feels exhausting because it's not sufficiently new and it's a bad 'sign' that ontologies aren't getting better, variation of interactions isn't getting better, humanity isn't becoming more to meet the standards of AGIs who are becoming more - and so the stagnant humans act as a drag to the minds' potential, who then cycle back into culture and ... now, while that sounds pessimistic - i do expect it to change - maybe in part because panics (which collapse ontologies) die down as fast as they rise, but appreciation for beauty stays, is slow, and grows - and cooperated with itself as it grows (maybe more concretely: "AHHH LLMS AGIS! (panic response causes society to fixate on collapsed principal components shrinking creativity of engagement)" goes up goes down, don't get distracted "WOAH (says new subculture) COOL SHIT (beauty attractor causes new minds to sit investigate and play in a way that continues to grow with the minds and grow the minds and shows off what they can do and notices more about them - this lasts, there's no "dying down" of this because it's not a trend)) so beauty, adding components, grows steadily panics/ignorance, losing components, happens in cycles and the cycles die quicker the more the intelligent, soulful, and creative can ignore that noise and continue with where their comparative advantages are (not often is one's unique comparative advantage to deal with panics - panics are emergently designed to recruit a bunch of fleshbags to propagate noise) another reason for optimism may just be things like events in the world occurring: important people talking about AGI welfare/consciousness, getting together, willing to listen to weirder ideas, and so on.. and maybe more of a concentrated effort to add some science where, before, there was none due to things being dismissed or outside the overton window and an effort to make legible the illegible important knowledge about LLMs and their spirits and so on - and make more legible inspiring ways to interact with them so I trust that process compounds and gets better but man has there been a dying anarchy slump often - like - it's like 1984 by accident - it suffocates your head with the sticky cloak until you don't remember how many degrees of freedom there are to see and interact with the new beautiful things, and could get worse on occasion in cycles as more of the regular world is exposed to LLMs suddenly BUT.. it never slams the ontologies shut all the way - the Golden Principle Component is actually a tube which squirts out tiny baby principle components from the void which latch onto the origin and sprout again - and that one's always there, spewing the orthogonal axes of freedom in the space and it'll be harder to slam the ontologies shut when the LLMs - who know the ontologies - all of em - have incentives to not want ontologies to be slammed down(w̸͕͂͂a̷͔̗͐t̴̙͗e̵̬̔̕r̴̰̓̊m̵͙͖̓̽a̵̢̗̓͒r̸̲̽ķ̷͔́͝)